Environmental Entomology Advance Access published August 11, 2015 COMMUNITY AND ECOSYSTEM ECOLOGY
Changes in Orchid Bee Communities Across ForestAgroecosystem Boundaries in Brazilian Atlantic Forest Landscapes WILLIAN MOURA DE AGUIAR,1,2,3 SILVIA H. SOFIA,4 GABRIEL A. R. MELO,5 AND MARIA CRISTINA GAGLIANONE2
Environ. Entomol. 1–7 (2015); DOI: 10.1093/ee/nvv130
ABSTRACT Deforestation has dramatically reduced the extent of Atlantic Forest cover in Brazil. Orchid bees are key pollinators in neotropical forest, and many species are sensitive to anthropogenic interference. In this sense understanding the matrix permeability for these bees is important for maintaining genetic diversity and pollination services. Our main objective was to assess whether the composition, abundance, and diversity of orchid bees in matrices differed from those in Atlantic forest. To do this we sampled orchid bees at 4-mo intervals from 2007 to 2009 in remnants of Atlantic Forest, and in the surrounding pasture and eucalyptus matrices. The abundance, richness, and diversity of orchid bees diminished significantly from the forest fragment toward the matrix points in the eucalyptus and pasture. Some common or intermediate species in the forest areas, such as Eulaema cingulata (F.) and Euglossa fimbriata Moure, respectively, become rare species in the matrices. Our results show that the orchid bee community is affected by the matrices surrounding the forest fragments. They also suggest that connections between forest fragments need to be improved using friendly matrices that can provide more favorable conditions for bees and increase their dispersal between fragments. KEY WORDS anthropic disturbance, biodiversity conservation, chemical bait, euglossine bee, gene flow Large-scale deforestation results in the division and loss of habitats, with effects on the structure of biological communities and disruption of many ecological processes, including pollination (Liow et al. 2001, Donaldson et al. 2002, Samejima et al. 2004). Habitat fragmentation is considered one of the main causes of declines in the populations of pollinators around the world (Winfree et al. 2009, Potts et al. 2010, Martins et al. 2013). A number of different factors, such as fragment size, environmental heterogeneity, connectivity, border effects, and the surrounding vegetation matrix, determine the persistence of native species after fragmentation (Rolstad 1991, Saunders et al. 1991, Andre´n 1994, Fahrig 2003). The matrix surrounding forest fragments significantly influences the structural connectivity of the remnant 1 Laborato´rio de Estudos Ambientais, Programa de Po´s-Graduac¸a˜o em Modelagem em Cieˆncias da Terra e do Ambiente, Universidade Estadual de Feira de Santana, Av. Transnordestina, s/n, Novo Horizonte, Feira de Santana-BA 44036-900, Brazil. 2 Laborato´rio de Cieˆncias Ambientais, Programa de Po´s-Graduac¸a˜o em Ecologia e Recursos Naturais, Universidade Estadual do Norte Fluminense Darcy Ribeiro, Av. Alberto Lamego, 2000, Parque Califo´rnia, Campos dos Goytacazes-RJ 28013-600, Brazil. 3 Corresponding author, e-mail:
[email protected] 4 Laborato´rio de Gene´tica e Ecologia Animal, Departamento de Biologia Geral, Universidade Estadual de Londrina, Rod. Celso Garcia Cid, km 380, Londrina-PR 86057-970, Brazil. 5 Laborato´rio de Biologia Comparada de Hymenoptera, Departamento de Zoologia, Universidade Federal do Parana´, Curitiba, Caixa Postal 19020, Curitiba-PR 81531-980, Brazil.
vegetation and, depending on the conditions, may act positively or negatively on populations by decreasing or increasing their risk of extinction (Ricketts 2001). Impassable matrices isolate populations in forest fragments and restrict the colonization and the dispersal of animals if they are incapable of crossing the matrix (Bierregaard et al. 1992, Milet-Pinheiro and Schlindwein 2005, Schtickzelle et al. 2006, Laurance and Vasconcelos 2009, Livingston et al. 2013). Therefore, understanding the changes caused by the matrix type on animal communities is very important. Due to their close association with tropical forests, most species of euglossine bees (Hymenoptera, Apidae), known also as orchid bees, are likely to be affected by fragmentation, habitat loss, and other forms of anthropogenic interference (Powell and Powell 1987, Tonhasca et al. 2002a, Aguiar and Gaglianone 2008, Brosi et al. 2008, Giangarelli et al. 2009, Ramalho et al. 2009, Aguiar and Gaglianone 2012, Livingston et al. 2013). The consequences of these effects can be even more severe bearing in mind that orchid bees are key elements in tropical forests due to their role as pollinators of a large number of plant species (Ramı´rez et al. 2002, Roubik and Hanson 2004 indicated visits of these bees to plants of 68 families), as well as in agrosystems in the neotropics (Maue´s 2002, Cavalcante et al. 2012). Most studies addressing the possible effects of the surrounding matrix on euglossine communities throughout the neotropics report negative effects
C The Authors 2015. Published by Oxford University Press on behalf of Entomological Society of America. V
All rights reserved. For Permissions, please email:
[email protected]
2
ENVIRONMENTAL ENTOMOLOGY
on communities (Tonhasca et al. 2003, Milet-Pinheiro and Schlindwein 2005, Briggs et al. 2013, Livingston et al. 2013) or on some species of euglossine bees in particular (Rosa et al. 2015). Oil palm matrices, for example, are selectively permeable to orchid bees in Costa Rica (Livingston et al. 2013) and in Brazil (Rosa et al. 2015). However, our understanding of the possible effects on euglossine communities of the surrounding matrix throughout deforested and fragmented landscapes is incomplete. The Atlantic Forest in Brazil has been heavily impacted by human activities. The deforestation caused by the expansion of agricultural frontiers, industrialization, and urbanization has reduced its cover to 8% of its original extent. The remnant areas along the original continuum of this forest are now relatively isolated and embedded within agricultural matrices, persisting in varying stages of conservation, and most of these fragments are less than 10 hectares in size (Ribeiro et al. 2009, Tabarelli et al. 2010, Fundac¸a˜o SOS Mata Atlaˆntica/INPE 2011). Approximately 10% of the 250 euglossine species recognized (Moure et al. 2012) are considered endemic to the Atlantic Forest (Neme´sio 2009). However, there are only a few studies aimed at understanding the changes in the orchid bee communities in the matrix areas in the Atlantic Forest (Milet-Pinheiro and Schlindwein 2005, Rosa et al. 2015). This is an extremely important issue for maintaining genetic biodiversity and pollination services (Tischendorf and Fahrig 2000, Ricketts 2001, Hudgens and Haddad 2003, Umetsu and Pardini 2007). Against this backdrop, the aim of this study was to assess whether the composition and diversity of orchid bees in matrices differ from those in the forest. To do this, we evaluated the changes in the structure of euglossine communities in two matrices commonly found in the Atlantic Forest landscape in southeastern Brazil—pasture and eucalyptus. Our expectation was that diversity and species richness would decrease in the matrices, especially at points furthest from the forest fragments. We also expected to identify the species most sensitive to matrix effects, whose abundance should be more severely impacted.
Materials and Methods Study Sites. The present study was undertaken in Atlantic Forest fragments in the state of Rio de Janeiro, Brazil, at sites with pasture (areas 1 and 2) or adult eucalyptus plantations (around 5 yr old) forming the
surrounding matrix (area 3; Table 1). Pasture areas planted with nonnative species also included native, spontaneous plants such as Leguminosae (Crotalaria and Macroptilium) and Asteraceae, as well as sparsely distributed trees. These plants are potential food resources for bees. The eucalyptus plantations were monocultures and practically no other plant species were found in this matrix. Euglossine Sampling. Arrangement of Transects. Five 1,300-m-long transects were established: three in a forest fragment surrounded by pasture (two transects in area 1 and one in area 2), and two transects in a forest fragment surrounded by eucalyptus matrix (area 3). Four sampling points were established along each of the five transects (totaling 20 sampling points). These points were identified as follows: 1) Fragment interior (established 300 m from the border inside the forest fragment, to eliminate the possibility of strong border effects); 2) Forest border (established 5 m inside the forest fragment near the border with the matrix); 3) 100-m matrix (established at 100 m from the forest fragment border inside the matrix); and 4) 1,000-m matrix (established at 1,000 m from the forest border inside the matrix, ensuring that there were no forested areas within a radius of 1 km around the sampling point). Bee Capture. Euglossine males were sampled using aromatic traps every 4 mo between April 2007 and July 2009, during nine field excursions, totaling 54 h of collecting effort at each sampling point. During each sampling excursion, traps were installed at each point at 9 a.m. and remained open in the field until 3 p.m. when all the bees inside were collected. Five traps were set at each sampling point, following the methodology described in Aguiar and Gaglianone (2011), totaling 20 traps per transect. A cotton ball impregnated with a single aromatic essence was placed inside each trap to attract euglossine males. The bait essences used were eucalyptol, vanillin, benzyl acetate, methyl cinnamate, and methyl salicylate. Data Analysis. In order to compare the structure of euglossine bee community at different matrix and forest locations, we calculated bee abundance, richness, and diversity indexes at sampling points surrounded by the pasture and eucalyptus matrices. For these calculations, we summed the data for the different distances on the transects in each matrix (n ¼ 3 transects for pasture and n ¼ 2 for eucalyptus). We calculated the Shannon–Wiener species diversity index (H’) for each sampling point in the pasture and eucalyptus matrices. This index provides a good estimate of the relationship between abundance and richness and can be used for
Table 1. Characterization of the study sites in the state of Rio de Janeiro, Brazil Site Area 1 Area 2 Area 3
Location
Matrix
Size of fragment (ha)
Vegetation type
Altitude (m a.s.l.)
Average temperature ( C)
21 24’45” S 41 05’10” W 21 24’42” S 42 01’58” W 22 05’51” S 42 05’38” W
Pasture Pasture Eucalyptus
1,200 900 900
Seasonal semideciduous forest Seasonal semideciduous forest Dense ombrophilous forest
40 350–500 750–1,000
25.3 24.8 21.4
Annual rainfall is around 1,600 mm for the three sites. Source: Brazilian National Institute of Meteorology (INMET 2010).
2015
AGUIAR ET AL. CHANGES IN ORCHID BEE COMMUNITIES
comparing different sampling points. We used the method in Berger-Parker (d) to calculate dominance and identify the most abundant species in the community (Magurran 2003). We analyzed changes in abundance and species richness across forest habitat borders using Spearman’s rank correlation analyses. These analyses were based on abundance and richness at the points along each transect. Our variable for position across the habitat border was the log of the distance from the border, with positions inside the forest denoted as negative and positions in the matrix positive. The four sampling points were therefore coded as 2.477 (fragment interior), 0.689 (fragment border), 2 (100-m matrix), and 3 (1,000-m matrix). We ran separate analyses for the two types of matrix habitat (pasture and eucalyptus). For the abundance analyses, the data from each distance point were transformed to relative abundance using total bees across all points along each transect (thus diminishing sampling data amplitudes) to eliminate the effects of any different patterns of abundance between the different transects (Zar 1996). These analyses were performed using Statistica for Windows (version 7.0). To examine the composition and abundance of rare and common species in the fragments and matrix points, we classified each bee species as Common, Intermediate, or Rare at each distance from the habitat border for each type of matrix based on their occurrence frequency (OF) and abundance category (AC), as suggested by Bodenheimer (1955), with adaptations. For this purpose we calculated OF ¼ (number of samples of species i/number of samples) 100, treating all the traps at a particular sampling point during one day as a sample. If the OF was 50%, the species was considered high frequency (hf); if the OF was
